Kaggle Challenge: Walmart stores sales forecasting

This repository presents a bunch of notebooks to tackle Kaggle Challenge i.e. Walmart Store sales forecasting.

Dataset

It consists of total 45 stores sales consisting of different products at different timestamps.

As it is a time series dataset, different forecasting algorithms were tried as presented in the notebooks. Forecasting algorithms are:

Linear Regression
Fb Prophet
LSTM (keras)
LSTM (Pytorch) For dimensionality reduction, auto-encoders were trained both in Keras and Pytorch.

Forecasting result of FP prophet model is:

Comparison of different models and approaches as compared to 1st ranked solution:

Name		Name	Last commit message	Last commit date
Latest commit History 5 Commits
Autoencoder+regression.ipynb		Autoencoder+regression.ipynb
Multivariable time series.ipynb		Multivariable time series.ipynb
README.md		README.md
fb_prophet result.PNG		fb_prophet result.PNG
fb_prophet_prediction.csv		fb_prophet_prediction.csv
fb_prophet_prediction.npy		fb_prophet_prediction.npy
out_AE.csv		out_AE.csv
regression.ipynb		regression.ipynb
rmse comparisons.PNG		rmse comparisons.PNG
time series keras.ipynb		time series keras.ipynb
time series pytorch.ipynb		time series pytorch.ipynb
time series.ipynb		time series.ipynb